AMD Radeon HD 6350M

The AMD Radeon HD 6350M is an end-of-life portable-device graphics processor from AMD. It uses the Robson chip and the TeraScale 2 architecture, fabricated by TSMC on a 40 nm process. The die contains 292 million transistors on 59 mm², giving a transistor density of 4.9M per mm². The product belongs to the Vancouver generation (HD 6300M), was released on 2010-11-25, and is positioned between the Manhattan predecessor and the London successor. The data does not list a codename or a series; the chip identifier is Robson.

Benchmark Performance

The benchmark list is empty; there are no synthetic or application scores. The average benchmark score is 0, and the percentile against all GPUs is 50. A percentile of 50 is the central position of a rank distribution, but because the average score is 0 and no nearest-rival entries exist, that percentile cannot be converted into a percentage lead or deficit versus any named GPU. The performance-related rates present in the data are FP32 throughput of 80.00 GFLOPS, pixel fill of 2.000 GPixel/s, and texture fill of 4.000 GTexel/s. These rates correspond to 80 shading units, 8 texture mapping units, and 4 ROPs. The memory subsystem is 1024 MB of GDDR3 on a 64-bit bus, with 12.80 GB/s bandwidth. The memory clock is 800 MHz, and the effective data rate is 1600 Mbps. No base clock, boost clock, or game clock is recorded, so dynamic frequency behavior is absent from this analysis. The FP16 field is also null, leaving only the FP32 figure as the floating-point metric. Without benchmark entries, these raw rates are the only quantitative performance profile in the data. The 64-bit bus width and 12.80 GB/s bandwidth are the only memory-path figures available, and the 1024 MB capacity is the only memory-size figure listed. The 800 MHz memory clock is the only physical clock in the record; the 1600 Mbps effective rate is the only data-rate figure. The 80.00 GFLOPS number is explicitly an FP32 rate, not a measured game score. The pixel and texture rates are exact resource limits, not aggregate results.

Ray Tracing and Feature Set

The rtCores field is null, and the tensorCores field is null; the data provides no dedicated unit counts for ray tracing or tensor operations. This means ray tracing performance cannot be quantified from this record. The underlying architecture is TeraScale 2, which is the pipeline context for the shader and fill-rate numbers. For software interfaces, the data lists DirectX 11.2 (11_0), OpenGL 4.4, and no Vulkan support. These API entries define the accessible feature set. Without a Vulkan entry, a Vulkan path cannot be assumed. In the absence of tensor cores and RT cores, any ray tracing or tensor workload, if exposed through the listed APIs, would have to use the general-purpose shader resources of the 80 shading units. The data does not include a separate ray tracing feature row; the only feature rows are the API levels, the architecture name, and the null RT/tensor fields. The API list is therefore the complete software compatibility statement for this product. The absence of Vulkan support is recorded directly in the data, while DirectX 11.2 (11_0) and OpenGL 4.4 are the only supported API versions listed.

Power and Cooling

The TDP is 7 W, the only power or thermal figure in the data. The suggested PSU field is null, and the power connector field is null; there is also no slot width or dimension data. The display output field is “Portable Device Dependent,” meaning the host system manages display connections. The bus interface is PCIe 2.0 x16. Memory timing is 800 MHz, with a 1600 Mbps effective rate. A 7 W TDP is a small thermal budget, but no cooler, heatsink, or system cooling requirement is listed. The production status is end-of-life, so no present-day cooling guidance is included in the record. No length, height, or width values are present, so physical mounting constraints are unspecified. The PCIe 2.0 x16 interface is the only connectivity standard listed. The display output field covers display integration, while power and cooling are left to the host platform because no PSU recommendation and no connector requirement are recorded. The 7 W figure is the sole thermal load indicator; with a null suggested PSU field, the data does not specify a power supply size.

How It Compares

The nearestRivals array is empty. There are no rival names, no rival scores, and no deltaPct values to cite. The only comparative field is the percentile against all GPUs, which is 50. Since the average benchmark score is 0, the percentile is a standalone rank value rather than a calculation based on a measured score. The predecessor field lists Manhattan, and the successor field lists London, but these are lineage entries, not benchmark rivals. Without nearest-rival data, relative statements such as “ahead of” or “behind” a specific card cannot be made. The empty benchmark list reinforces that no score delta can be computed. Therefore, the only supported comparative claim is the 50th-percentile field, with no other GPU-to-GPU numeric comparison available. The data’s nearest-rival list contains no objects at all, so there are no deltaPct percentages to analyze. The Manhattan and London entries are generation context, not score context. The average benchmark score of 0 and empty benchmark array together mean that any comparison would need external sources beyond this record.

Who Should Consider It

The Radeon HD 6350M is an end-of-life product released on 2010-11-25, and there are no benchmark scores to anchor a resolution- or settings-level recommendation. The resource envelope is 80 shading units, 8 TMUs, 4 ROPs, and 1024 MB of GDDR3 on a 64-bit bus. Bandwidth is 12.80 GB/s, FP32 throughput is 80.00 GFLOPS, pixel rate is 2.000 GPixel/s, and texture rate is 4.000 GTexel/s. Any workload must fit inside that memory-bandwidth and fill-rate budget. The 40 nm TeraScale 2 architecture, DirectX 11.2 (11_0), OpenGL 4.4, and absent Vulkan support define its software generation. The PCIe 2.0 x16 interface is the host connection, and the “Portable Device Dependent” display output means the host provides display ports. With a 7 W TDP, the component is low-power, but low power is not measured performance. The 1024 MB memory capacity and 12.80 GB/s bandwidth are hard limits for any application. The data supports no claim about specific resolutions, quality presets, or game titles. This is a legacy, end-of-life part whose position is defined by raw resource counts and a 50th-percentile field, not by an average benchmark score of 0. The transistor count of 292 million, die size of 59 mm², and transistor density of 4.9M per mm² are manufacturing facts rather than performance indicators, but they describe the physical implementation that hosts the 80 shading units, 8 TMUs, and 4 ROPs. The memory type is GDDR3, the memory size is 1024 MB, and the bus width is 64 bit; these are the fixed storage and transfer parameters. The API list is DirectX 11.2 (11_0) and OpenGL 4.4, with Vulkan not recorded. The release date and end-of-life status place this GPU in a specific historical generation. For any user, the practical boundary is the same: 80.00 GFLOPS, 2.000 GPixel/s, 4.000 GTexel/s, 12.80 GB/s, and 1024 MB are the data-defined limits, and no benchmark result exists to demonstrate behavior beyond those limits.